Magnetic Resonance Imaging

In Magnetic resonance imaging (MRI, formerly referred to as Nuclear Magnetic Resonance imaging), the study object is placed within a high intensity magnetic field. This causes the magnetic moments of the molecules within the object to become aligned. The object is then irradiated with pulses of low-level microwave radiation (excitation pulses) that cause some of the magnetic moments of the molecules to oscillate and re-emit microwaves after each pulse. These re-emissions are measured and stored digitally. By introducing gradients in the background magnetic field, it is possible to determine the spatial location of a re-emitted microwave. An image representing various characteristics about the molecular emissions at discrete samples throughout the object scanned is then formed. By modifying the frequency and timing characteristics of the excitation pulse, and the delay time before measurement of the emitted energy, it is possible to image particular types of molecules (water for instance), movement (blood flow), and many other characteristics.

The output of an MRI scanner is similar to CT. Slices representing slabs of the object scanned are produced. However, unlike CT which always produces transaxial slices, the slices from MRI can be oriented in any plane. Additionally, it is quite common to scan the patient with different parameters in order to show different characteristics. Thus, it is common to obtain 2 or 3 sets of slices for one patient. Depending upon the type of acquisition, the multiple acquisitions may be registered spatially (because the same excitation pulse is used for multiple measurements) or not (if the studies were performed sequentially).

The output values at each image element are not calibrated to any particular scale. Generally they are 10 bit data samples. The values will vary depending upon the scan parameters, and the patient's size and magnetic characteristics. Additionally, the values are not constant over the entire scan space as inhomogeneity in the magnetic field causes pixels that may represent the same tissue, but located some distance apart to give different signals. This lack of an absolute scale for a dataset is a cause of much consternation to the researcher attempting to segment the MRI data. Modern MRI scanners can generally acquire a set of slices (32 or so) within five minutes. An entire study of a patient generally represents two to three sets of slices, with a study time of 30-45 minutes. Each slice usually represents a thickness of 2-10mm and contains 256 by 256 pixels. Each pixel represents from 1-5mm. The radiation dose from an MRI scan is negligible.


back.gif (279 bytes) Medical Imaging Techniques
 HyperVis Table of Contents

Last modified on March 02, 1999, G. Scott Owen, owen@siggraph.org